Everything in this program has been training you to be an active investor: to analyze businesses and pick them, betting you can do better than owning the whole market. That bet deserves the most rigorous, evenhanded scrutiny you can give it, because the evidence against it is strong and honest people land on both sides. Here is the debate in full, so you can judge for yourself whether your own edge justifies the game.
What is the efficient market hypothesis?
The Efficient Market Hypothesis, in three forms (Fama 1970). The EMH says asset prices already reflect available information, so you can't systematically beat the market using that information. It comes in three strengths:
- Weak form: prices already reflect all past price and volume information. The evidence supports it in a specific and limited shape, and it is worth stating precisely rather than as a slogan: simple price-only trading rules, the named chart patterns and the oscillator thresholds, do not survive transaction costs, and most of them have never been specified tightly enough to be tested at all. The documented exception is momentum, the tendency of three-to-twelve-month winners to keep winning, which is built from past prices alone and is one of the best-replicated anomalies in finance (Jegadeesh & Titman 1993). A price-only signal that survives costs across markets and decades is a weak-form violation by construction, so hold the weak form as "past prices are nearly exhausted, with one large and well-documented exception" rather than as "past prices are worthless." M9.05 takes the whole of technical analysis on those terms, evidence first and chart lore second.
- Semi-strong form: prices reflect all publicly available information: past prices plus financial statements, news, filings, everything public. If true, then fundamental analysis of public information, the entire craft this program teaches, cannot produce excess returns on average, because the market has already priced in what the 10-K says the moment it's filed. This is the form that matters most, and it is genuinely contested.
- Strong form: prices reflect all information, public and private. If true, even corporate insiders trading on secret information couldn't beat the market. This form is clearly false: insider trading demonstrably works, which is exactly why it's illegal (M0.02, the ethics thread). Nobody serious defends strong-form efficiency.
Advertisement
What does the evidence say about beating the market?
The evidence for the passive side (it is strong). Two independent lines of argument make the case that most active management is a losing game after costs:
- The arithmetic of active management (Sharpe 1991). This is not an empirical claim but a mathematical identity: before costs, the return of the average actively-managed dollar must equal the return of the average passively-managed dollar, because together they are the market. Therefore, after costs, the average active dollar must underperform the average passive dollar by exactly the difference in costs. Active management as a whole cannot beat passive as a whole after fees. It is arithmetically impossible. The only question is whether you specifically are in the winning minority, and the average participant, by definition, is not.
- The empirical record (SPIVA and kin). S&P's SPIVA scorecards, updated twice a year, consistently show that over long horizons the large majority of active funds underperform their benchmark after fees: on the order of ~85–90% of US large-cap active funds underperforming the S&P 500 over the 15 years to the early 2020s (as of mid-2026; pull the current SPIVA report, the exact figure moves). India's SPIVA tells a similar story for large-caps, where the majority of active funds lag over long horizons, though Indian mid- and small-cap active funds have historically fared somewhat better against their benchmarks than US large-cap active has (a hint about where inefficiency lives; check the current data). Persistence studies add the knife: the few funds that outperform in one period rarely repeat, which is what you'd expect if most outperformance is luck.
The Grossman-Stiglitz paradox (the case that markets can't be perfectly efficient). Now the counterweight, and it is a beautiful piece of logic (Grossman & Stiglitz 1980, "On the Impossibility of Informationally Efficient Markets"). Suppose markets were perfectly semi-strong efficient, with every price instantly reflecting all public information. Then no one would be paid for gathering and analyzing information, because you couldn't profit from it (it's already in the price). But if no one gathers information, how does it get into prices? It can't. Therefore perfectly efficient prices are self-contradictory: markets must be inefficient enough to compensate the people who do the analysis that makes them efficient. This is often called an "equilibrium degree of disequilibrium": there must be enough mispricing to pay the informed for correcting it. The paradox doesn't say active management is easy; it says active management must be rewarded at the margin, for someone, or the whole system breaks. The reward accrues to the best-informed, which raises the only question that matters for you: are you among them, on the specific bets you make?
Which market anomalies are real?
An anomaly is a documented, repeatable pattern in returns that the efficient-market hypothesis says should not be there. They are the empirical middle of the argument above, and an analyst is expected to be able to name them, sort them and say how much weight each carries. The standard sort has three families. Time-series anomalies claim a pattern in an asset's own returns across time. Cross-sectional anomalies claim that a characteristic separates winners from losers at a point in time. The rest are structural oddities that fit neither.
| Family | Anomaly | The claim | Where the evidence stands |
|---|---|---|---|
| Time-series | Calendar effects | Returns cluster by month, weekday or turn of the month (the January effect, the weekend effect) | Weak and fading. Mostly small-cap and micro-cap, largely gone since publication, and rarely bigger than the spread you cross to trade it |
| Time-series | Momentum | Three-to-twelve-month winners keep winning for a few months | Strong and replicated across markets and asset classes (Jegadeesh & Titman 1993). Survives costs at institutional scale, and crashes savagely at turning points (see factor investing) |
| Cross-sectional | Size | Small caps out-return large caps for their beta | Real but much weaker than first reported. Concentrated in micro-caps and January, and heavily reduced once illiquidity and quality are controlled for |
| Cross-sectional | Value | Cheap-on-book or cheap-on-earnings out-returns expensive | Long history, then a drought from roughly 2007 to 2020 severe enough to be a live argument about whether the premium is intact |
| Cross-sectional | Low volatility | Low-beta and low-volatility stocks earn more per unit of risk than CAPM allows | Persistent and widely replicated. The leading explanation is a leverage constraint that pushes constrained investors into high-beta names (Frazzini & Pedersen 2014) |
| Other | Post-earnings-announcement drift | Prices keep drifting in the direction of an earnings surprise for weeks after the print | One of the oldest and most robust. Also one of the most costly to harvest, because it needs fast, high-turnover implementation |
| Other | Closed-end fund discount | Identical assets trade below net asset value inside a listed wrapper, and the discount varies | The textbook demonstration that price and value can separate for a portfolio whose value is not in dispute. India's version sits in listed holding companies (M3.10) |
| Other | Overreaction | Three-to-five-year losers beat three-to-five-year winners over the following three to five years | Documented (De Bondt & Thaler 1985) and directionally the opposite of momentum, which is why horizon is not a detail here. Sensitive to how risk is measured |
Two explanations account for almost every entry in that table, and they are not mutually exclusive. The first is risk mismeasurement: the pattern is a real return, but it is compensation for a risk the model failed to price, so what looks like free money is a fee you are being paid for bearing something CAPM cannot see. Value and size are usually argued this way. The second is data mining with post-publication decay: with enough researchers testing enough characteristics on the same few decades of data, some patterns are the survivors of a large hidden search rather than facts about the world, and the test is what happens after publication. The measured pattern is that anomaly returns fall materially once a paper appears, partly because the finding was overfitted and partly because capital arrives to trade it. Calendar effects are the clearest casualty.
The working rule for a bottom-up investor is therefore narrower than the table looks. An anomaly you cannot implement after costs is a fact about a research dataset, not an opportunity, and the two most robust entries above are also two of the most expensive to trade. Treat the table as a map of where prices are known to be sticky, and then ask the factor tilt question about your own book: if your returns are the anomaly rather than your analysis, the anomaly can be rented. AA1.07 formalises what remains after that subtraction, with active share, active risk and the fundamental law of active management; AA1.08 covers what happens when you decide to rent it instead, through index construction, replication and tracking error.
Where can an active edge persist?
Where an active edge plausibly persists. Efficiency is not uniform; it is thin in some corners and thick in others. The honest map of where edge is most defensible:
- Small and micro-caps. Less analyst coverage, less institutional ownership, more genuine mispricing. The efficiency thins where the money and the eyeballs are absent.
- Special situations. Spin-offs, restructurings, distressed debt, complex holdcos (M3.10, Greenblatt in M6.02). Forced and indiscriminate selling (index funds dumping a spin-off), complexity, and small size create mispricing that patient analysis can exploit.
- Time-horizon arbitrage. The single most durable and scalable edge available to an individual. Most market participants are forced to care about the next quarter (career risk, redemptions, benchmarking). An investor who can genuinely hold for five years is exploiting a different asset than the one the quarterly crowd is trading, and that patience is an edge no amount of information competes away. It is the one edge that gets stronger, not weaker, as markets get more short-term.
- Less-efficient markets and pockets. India's mid- and small-caps, promoter-heavy names where governance complexity scares off lazy capital, illiquid situations, and moments of forced selling or euphoria. Emerging-market inefficiency is real, though it comes bundled with real governance and fraud risk (the ethics thread) that is the reason the inefficiency exists and the reason it demands deeper diligence.
- Behavioral mispricings. The systematic errors of factor investing and M6.04 (overreaction, neglect, narrative-chasing) that create the value and momentum premia in the first place. Where humans are predictably irrational, a disciplined process has an edge.
Should you invest actively or passively?
The honest conclusion: let the learner judge, but here is where the weight of the evidence points. Passive indexing is the correct default for most investors most of the time: it is arithmetically guaranteed to beat the average active dollar after fees, it removes idiosyncratic risk for free, it demands no skill you may not have, and it protects you from your own behavioral errors. Anyone without a real, identified, durable edge should index, and there is no shame in it. It is the intellectually honest choice. Active management is justified only when you can name your specific edge (which corner, which mispricing, why it persists, why you can exploit it) and can honestly show, benchmarked against the factor blend that matches your style (see factor investing) rather than against the broad index, that it is real. The purpose of this entire program is to give you a genuine shot at being in the justified minority: to build the analytical depth, the temperament, and the process that constitute a real edge in the corners where edge persists, and, just as important, the honesty to know if you don't have one, and to index the capital for which you don't. The learner who finishes this program and concludes "I'll index my core and run a concentrated active sleeve only in Indian mid-caps and special situations where I've shown an edge" has understood the argument perfectly. E11.05 turns exactly that sentence into a written policy, with the index default stated as the position you have to argue your way out of rather than the one you fall back on.